Open-Source Tool Kit Uses 3D Printing for Micromodel Generation

You have access to this full article to experience the outstanding content available to SPE members and JPT subscribers.

To ensure continued access to JPT's content, please Sign In, JOIN SPE, or Subscribe to JPT

In this paper, the authors present an open-source tool kit for the generation of microfabricated transparent models of porous media (micromodels) from image data sets using optically transparent 3D polymer additive manufacturing (3D printing or sintering). These micromodels serve as research and pedagogical tools that facilitate the direct visualization of drainage and imbibition within quasi-2D porous media, generated from a range of image modalities [e.g., thin section micrographs, µ-computed tomography (μCT) orthoslices, and conventional digital photography].


Though recent advances in 3D X-ray imaging, such as X-ray microtomography and µCT, permit volumetric imaging of microcore flood experiments within geological samples at the pore scale, experimental observations of dynamic (time-resolved) multiphase flow within pore networks still are obtained conventionally using transparent etched or molded synthetic porous media commonly referred to as micromodels. Typically, video footage of fluid imbibition and drainage experiments conducted across these quasi-2D pore networks is used to understand fluid distributions and displacement mechanisms within an equivalent 3D porous media. Contrary to state-of-the-art dynamic µCT coreflood experiments, which require synchrotron beam time to be conducted, micromodel studies can be undertaken routinely within a laboratory-based setting with a relatively simple experimental setup. However, the facilities required to fabricate micromodels typically are highly specialized, with production often outsourced to third-party manufacturers.

In this work, the authors consider the potential of using additive manufacturing (3D printing or sintering) as an alternative to conventional micromodel-fabrication techniques. With the advent of light-transmissible 3D-printable materials, flow experiments conducted with these 3D-printed physical models can be captured using widely available optical imaging techniques (i.e., standalone digital cameras and trinocular microscopes). However, a major obstacle encountered in harnessing 3D-printing technologies for the goal of micromodels fabrication is the paucity in software tools capable of processing and converting raster-based images to mesh-based file formats parsable by commercially available 3D-printing systems.

The paper aims to provide an open-source platform through which 3D-­printable mesh-based representations of micromodels can be generated from raster imagery. Using this tool, standard micromodel components are fully integrated into the fabricated model. The availability of such a tool kit could act as an enabler for community research into fluid-transport phenomena in porous media. The authors suggest that 3D-printed pore networks may provide an effective pedagogical tool for teaching multiphase flow, enabling petroleum- and chemical-engineering and geology students to visualize directly often obtuse immiscible-fluid-flow processes within the classroom.

Software-Implementation Description

The tool kit is developed in the popular MATLAB language and is executed by graphical user interface. The generation of 3D-printable micromodels from raster datasets comprises three main stages:

  1. Image processing
  2. Micromodel construction
  3. Mesh post-processing

Image Processing. Raster (2D) imagery forms the basic input data required for micromodel generation. Imagery can be sourced from a variety of imaging modalities. To be used for micromodel generation, three-channel red-, green-, blue-, and gray-scale images must be segmented into binary images. The presence of noise in image data is detrimental to the quality of the output of most commonly applied segmentation routines. To reduce noise before segmentation, the authors provide an implementation of an edge-preserving nonlocal-means filter detailed elsewhere in the literature, which has significant advantages over window-averaging-based methods in terms of preserving contrast at the interface between different phases. For segmentation of input raster data into binary images, both gradient-thresholding and watershed-transform-based methods are provided, with cropping tools made available to enable user-specified regions of interest to be used for micromodel generation.

The cleaning of artifacts attributable to image segmentation, such as holes, isolated pixels, and small pixel clusters, before micromodel generation is desirable. The tool kit contains a number of binary morphological operations used for cleaning image artifacts, including small spot and hole removal, dilation and erosion, and morphological opening and closing. In addition, because of the presence of grain-to-grain contacts and the loss of pore throats as a result of artifact cleaning, 2D image data of real rock pores (especially thin-section photomicrographs and µCT orthoslices) tend to exhibit lower effective porosities than their volumetric counterparts (e.g., µCT voxel images). Reinstatement of pore throats between noncommunicative pores is often necessary to use such images in micromodel-based flow experiments. This is achieved through use of the authors’ software tool by computer-aided-design-based tools, which can be used to manually cut pathways between merged grains.

Micromodel Construction. Synthetic porous media generated using the tool kit contain standard micromodel components (i.e., inlet and outlet chambers, micromodel perimeter walls, cover slip, and pore network) integrated into the fabricated model. The tool kit also produces guide holes for the location of inlet ports, which must be installed post-fabrication by the user, a task best accomplished using a pillar drill and tap wrench. Typically, microfluidic flowlines and flangeless connectors with standardized metric or imperial screw threads are used to connect injection and production syringe pumps to the micromodel assembly.

Integrated micromodel components are constructed around the selected binarized porous media image in a voxel space, the dimensions of which are defined by the user. Having generated this initial model, voxelized components are tessellated to form a mesh-based representation of the micromodel, which can be parsed by most commercially available 3D-printing systems.

Mesh Post-Processing. Tessellation of binary voxels results in blocky mesh surface geometries, introducing roughness artifacts into the output porous media. The authors have implemented Laplacian mesh filters in order to remove surface-roughness artifacts arising from the mesh-generation process. Finally, because of the occurrence of bottlenecks in the processing of large triangular irregular network models by commercially available 3D printers, reducing the face count of 3D meshes before their use as a template for additive-manufacturing-based fabrication is often necessary. The authors therefore provide surface-preserving mesh-decimation functions to reduce output models to data volumes that can be readily parsed by most 3D printers.  

Suggested Experimental Procedure

The experimental procedure for the study of multiphase flow using 3D-printed micromodels is equivalent to that used with conventional etched or molded models of porous media. The authors provide a prospective design for n-decane-water drainage and imbibition experiments using 3D-printed micromodels generated using the open-source tool kit (Fig. 1). Micromodels of Berea sandstone are fabricated for installation on the stage of an inverted trinocular microscope, operated in fluorescence mode and using a 2X objective. Fluids are doped with an ultraviolet-sensitive contrast agent. Injection of fluids is conducted under ambient pressures and temperatures using syringe pumps. Under such an experimental setup, both primary drainage and imbibition can be visualized and analyzed by time-series image data, enabling both irreducible water saturation and residual oil saturation to be quantified.

Fig. 1—Proposed experimental setup.


The authors believe that providing researchers with the capability to prototype and fabricate micromodels with different configurations rapidly and at variable-length scales may act as a major enabler within porous-media studies, removing cost and time constraints incurred by conventional manufacturing techniques. In addition to effectively limitless configuration options in terms of pore-network geometry and injection-production port patterns, this tool kit permits the assessment of a variety of materials applicable to additive manufacturing, providing workers with increased flexibility in experimental design. Moreover, the models produced can be applied to more-conventional fabrication techniques (i.e., computer-numerical-control-based etching or milling systems), further extending the utility of the application.

It should be noted that, while feasible, the production of optically transparent objects using currently available 3D-printing systems remains challenging. A common issue with the most widely available additive manufacturing technology (i.e., fused depositional modeling) is that the pores between consecutively deposited layers result in the diffraction of light through the printed object, producing a translucent, milky finish. Thus, many currently available lower-cost 3D-printing systems are potentially unsuitable for transparent micromodel production. A further limitation is the XY resolution of many 3D-printing systems, which is commonly greater than 100 µm. The capacity of such systems to reproduce realistic pore geometries in fine-grained geologic media (e.g., coarse-grained sands and below) is therefore severely limited. However, developments in direct-laser-­writing-based 3D lithography now enable intricate nanoscale models to be fabricated, holding great promise for the realistic representation of the micro- to nanoporous domain in micromodel experiments.

The authors believe that, given rapid and ongoing development, additive-manufacturing technologies hold significant potential for micromodel-based studies into transport phenomena within porous media. It is hoped that the tool kit will act as an enabler to researchers, allowing far greater flexibility in the design of micromodel experiments than can be realistically achieved using traditional fabrication techniques. Fundamental insights gained by such research will contribute to the development of more-effective strategies for maximizing oil and gas recovery within a range of reservoir settings.

This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 190852, “Open-Source Tool Kit for Micromodel Generation Using 3D Printing,” by Thomas D. Seers and Nayef Alyafei, SPE, Texas A&M University at Qatar, prepared for the 2018 SPE Europec featured at the 80th EAGE Conference and Exhibition, Copenhagen, Denmark, 11–14 June. The paper has not been peer reviewed.

Open-Source Tool Kit Uses 3D Printing for Micromodel Generation

01 July 2019

Volume: 71 | Issue: 7